AI Powered Contact Tracing System with Risk Assessment

Abstract

The challenges posed by the COVID-19 pandemic have demonstrated the critical importance of having robust contact tracing systems to enable the early detection of infection spread and the scaling of risk response initiatives. Manual contact-tracing systems are cumbersome, prone to error, and compromise user privacy. Given the limitations highlighted, this paper presents a novel AI-driven contact tracing system that combines machine learning, edge computing, and privacy-preserving mechanisms to enhance accuracy and reduce response lag times. The developed system relies on wireless Bluetooth and GPS data from smartphones to compute proximity events between subscribers. An AI risk assessment model processes the events to compute infection potential based on encounter duration, environmental interaction, and personal health history. A comparative experimental simulation demonstrated that the model could achieve 93% accuracy in identifying a high exposure risk while still protecting FERPA principles through federated learning. Therefore, AI-based systems offer a viable option for scaling contact tracing systems.